Practical Codex Workflow: Saving Tokens and Preventing AI Laziness
dotey · x · 2026-08-02
Discusses how to effectively save tokens and maintain context quality when using coding agents like Codex for continuous tasks.
- Context Management: With improvements in the model's native context compression, the previously popular practice of manually creating a 'handoff' to start a new session is often no longer necessary; simply using /compact is usually enough. However, handoffs remain useful for cross-agent collaboration (e.g., passing a session from Claude Code to Codex).
- Workflow Optimization: The author prefers drafting and refining a technical design doc using Fable 5 first, then handing the document to Codex along with a /goal prompt to drive development.
- Anti-Laziness Mechanism: Emphasizes the necessity of setting strict acceptance criteria for tasks. For instance, in a code migration task, vague instructions lead to mediocre results. But if explicitly instructed that 'the UI pixels must perfectly match the original,' the agent will automatically take screenshots at each step to compare pixel differences until the standard is met.
Related event: Managing AI Coding Context: No Need to Overworry About Long Context(3 posts)→
More from coding & agent
- DeepMind turns coding agent Antigravity into a science tool, surfacing novel antimicrobial peptide in one session — pushmeet · 2026-10-16
- Cross-review coding workflow: Claude writes the plan, Codex audits it — theteknosaur · 2026-10-16
- Simon Willison: code mode is the future of tool calling, if it can be sandboxed — springrod · 2026-10-16
- Normalizing Claude's schema-free MCP output before it hits downstream n8n nodes — snowingbol · 2026-10-16
- Agent named 'Ilya' reportedly scores 1.1pt higher on benchmarks — daniel_mac8 · 2026-10-16
- A plain notes.md fixed my agent's cross-session amnesia better than chat history — rahul_watertech · 2026-10-16